From: Egmont K. <eg...@uh...> - 2006-06-02 16:37:06
|
On Fri, Jun 02, 2006 at 11:10:29AM -0400, Joseph H Allen wrote: > The characters to left of the <> should be in the 8-bit version of the > locale. The characters to the right of the <> should be in UTF-8. Brrrrrrrrr. Sorry but this sounds terrible. Joe already performs plenty of charset conversions, even when reading .po files (based on their header), I can't see why it couldn't do it here. Translators are usually not developers. They just sit at their properly configured system and use some tool (e.g. joe, kbabel...) and type the accented letters from their keyboard. Usually they are unable to "hack" in these kinds of various encodings into a file. They rightly expect things "just work", that is, if they type a particular accented letter, it should always remain the same and always mean that same letter. Tools such as kbabel, msgfmt and so on may also easily fail if a .po file claims it is UTF-8 while it isn't actually. They may fail to load or handle these files. And porting joe to real gettext will be harder, too. Furthermore, there might be languages with more old-fashioned locales, e.g. ru_RU.KOI8-R vs. ru_RU.ISO-8859-5. -- Egmont |